We wish to approach the direct analysis of the human genome in terms of comparisons between DNA sequences of test and reference sources. Our immediate concern is with the detection, resolution, and quantitation of restriction enzyme fragments of DNA and the isolation of fragments that are not common to both the test and reference sources. We hope to detect the presence, absence, or heterozygosity of specific fragments by survey procedures which will examine a large part of the genome. We have recently developed a high resolution separation of double helical fragments according to properties of their nucleotide sequence by electrophoresis in a gel containing a gradient of denaturants. Since the separation is nearly independent of the length of the fragments, we have combined it with conventional electrophoretic velocity separations based on length to give a two-dimensional array of well resolved spots by which at least 103 fragments can be recognized and isolated. We wish to refine and continue development of this technique and establish its quantitative capabilities. We hope to apply computational procedures for pattern recognition to the comparison of the E. coli chromosome with and without integrated lambda phage DNA, where we have already demonstrated discrete pattern differences as a step toward fully objective array comparisons among other organisms. We will apply transfer/hybirdization methods to identify points in our array with defineable biological functions and to establish the chromosomal sequence of separate fragments, at least in part. We will be concerned initially with viral and bacterial DNA. We hope to extend these techniques as early as possible to the analysis of eukaryotic systems, beginning with deletions and mutations affecting restriction enzyme fragmentation in yeast. A closer approach to the human genome will be provided by study of a series of deletion alleles at the c (albino) locus of chromosome 7 of the mouse. We will undertake dense procedures to isolate the DNA fragments that are present in wild-type mouse genome but not in the deletion alleles.